Exercise improved sleep in mental health patients

Disturbed sleep travels with mental health conditions so reliably that it is often treated as part of the illness rather than a target in its own right. Exercise has been proposed as an evidence-based complementary approach for sleep problems across many populations, but clinical psychiatric groups are frequently studied for mood outcomes rather than sleep. If exercise reliably improves sleep quality here, the follow-up question is which characteristics of a programme matter: how often, how hard, how long each session, and what type. This review set out to estimate both the overall effect and those programme details.

The authors searched MEDLINE, Embase, the Cochrane Library and Web of Science from inception to 1 March 2026 for randomised controlled trials of exercise interventions in patients diagnosed with psychiatric disorders, with sleep quality as an outcome. Risk of bias was assessed using the revised Cochrane Risk of Bias Tool 2. Effects were pooled as Hedges' g using random-effects models, giving 21 randomised trials contributing 24 comparisons and 1628 participants. Subgroup analyses were organised around the FITT framework of frequency, intensity, time and type, and univariate meta-regression tested duration, frequency, volume and total weekly exercise time as moderators.

Overall, exercise improved sleep quality with a moderate pooled effect of g = -0.57 (95% CI -0.75 to -0.38). Subgroup estimates were larger for programmes run at least three times a week (g = -0.60), for aerobic modalities (g = -0.69) and for low-to-moderate intensity work (g = -0.70). Sessions of at least 30 minutes (g = -0.58) and total programmes shorter than 12 weeks (g = -0.58) also showed larger effects. Crucially, meta-regression found no significant linear moderating effect of intervention duration, frequency, volume or total weekly exercise time, all with P above 0.05.

This aligns with a wider literature suggesting exercise helps sleep across a range of groups, and adds a pooled estimate specific to psychiatric populations. The tension inside the paper is instructive: the subgroup comparisons hint that moderate aerobic work done regularly performs best, yet the meta-regression found no continuous dose-response, so those patterns should not be read as an optimal prescription. Sleep quality in these trials is typically captured by self-report questionnaires, which are sensitive to expectation effects in interventions that cannot be blinded. Whether the benefit persists after the programme ends is not addressed.

The reasonable takeaway is that exercise is a credible, non-drug lever on sleep quality for people living with mental health conditions, with an effect size in the moderate range rather than a trivial one. The data lean towards regular, moderate aerobic activity rather than exhaustive high-intensity work, but the absence of a dose-response signal means no specific formula is established by this evidence. Because the trials involved diagnosed patients, the size of the benefit for an otherwise healthy person sleeping badly is uncertain. It is a reason to think movement and sleep are linked in both directions, not a protocol.